strisunshinewentingwang/pytorch

Pull Requests Commits

[FX] Make `immutable_list` and `immutable_dict` work with pytrees (#73766)

James Reed committed 2 years ago

a8d9fbb0

[FX lowering] Modify replace_all_uses_with to allowing filtering of nodes to update; use it to (#73763)

jaybean-dev committed 2 years ago

5332d870

Fix `functional.max_poolNd` warning spam in the CI

krshrimali committed 2 years ago

452c26bb

[quant][fx] Refactor _convert_fx_do_not_use (#73777)

jerryzh168 committed 2 years ago

8b8fac91

[Model Averaging] Support hierarchical model averaging (#73285)

wayi1 committed 2 years ago

0bb3b065

[torch.distributed][DDP] Disable DDP bucketing for the first iteration (#72843)

xush6528 committed 2 years ago

bcd0843b

dbr quant: enable reference module support for torch.qint32 (#73493)

vkuzo committed 2 years ago

727debb1

dbr quant: insert activation obs explicitly, instead of relying on hooks (#73492)

vkuzo committed 2 years ago

5787a36e

Automated submodule update: FBGEMM (#73277)

facebook-github-bot committed 2 years ago

350ecc9d

[AutoAccept][Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

CodemodService FBSourceClangFormatLinterBot committed 2 years ago

6067054c

Revert D34599476: [Quant][test] Added test to check if fp16 packing->unpacking yields the same result as to(torch.float16).to(torch.float32)

albanD committed 2 years ago

8a47d9cf

[NS] Mark output logger impure to avoid being removed in acc tracer (#73745)

Haixin Liu committed 2 years ago

3042f0ce

Update fbcode symlinks for mkl-dnn ideep 2.5.2

Wei Wei committed 2 years ago

bd3db019

[quant][fx] Add lowering support for qat and fused convs (#73527)

jerryzh168 committed 2 years ago

f5c7e540

[Quant][fx] Added explicit entries for for functional and module conv&linear support into get_default_qconfig_dict&get_default_qat_qconfig_dict (#73528)

dzdang committed 2 years ago

a39e8e8f

Eliminate unused parameters in PyTorch (#73749)

r-barnes committed 2 years ago

72e4aab7

Eliminate unused parameters in PyTorch (#73751)

r-barnes committed 2 years ago

a7d2f39b

Don't discard stacktrace when rewriting AttributeError (#73720)

ezyang committed 2 years ago

c24783fb

Add `.jenkins/caffe2/*` to ONNX merge rules

malfet committed 2 years ago

0642285e

[Quant][core][devs] Removed support for non-quantized tensors in reflection_pad1d_cpu (#72485)

dzdang committed 2 years ago

a4160223

GHF: Add clickable links for Approved By

seemethere committed 2 years ago

0b79715d

[Quant][test] Added test to check if fp16 packing->unpacking yields the same result as to(torch.float16).to(torch.float32) (#73685)

dzdang committed 2 years ago

37d1ec9d

[SR] Fix a kwargs API default value bug (#73681)

Mike Iovine committed 2 years ago

818bf361

[static runtime] Fix aten::relu TE lowering to enable vectorization (#73713)

huiguoo committed 2 years ago

75f5ceaa

Refactor tools/actions_local_runner.py to allow custom remote/branch names

Thiago Crepaldi committed 2 years ago

cf143bdc

[shard] fix init_from_local_shards issue with deepcopy (#73400)

wanchaol committed 2 years ago

b955a046

Improve Python 3.10 compatibility

ezyang committed 2 years ago

22a3a17e

[Core ML] Attemp to fix the OOM issue (#73750)

xta0 committed 2 years ago

e6f8e521

Add a default implementation of __torch_dispatch__

ezyang committed 2 years ago

35cfa74f

[ci] don't build pytorch twice on windows

suo committed 2 years ago

7c0f166b

Newer Older